The Triconch Sanctuaries of Sohag
“Le probleme de I’abside treflee, de son origine, de son developpement, I’histoire enfin de ce type de structure, a ete dans ces dernieres annees 1’argument de bien de travaux; mais, malgre cela, je crois que l’etude peut donner encore quelque resultat si on voudra profiter d’une documentation bien precise et abandonner toutes les theories aprioristiques.”  Published in 1925, these words still describe the situation nearly a century later: despite much recent discussion about the origin and history of the triconch sanctuary (‘abside treflee’), there is still much we do not know, and new research continues to challenge old theories and assumptions. This paper will review the course of scholarship since Ugo Monneret de Villard wrote those sentences, especially as it pertains to the churches of the monasteries of St. Shenoute and St. Bishoi.
The triconch—trichorum or toiko’/’/oc—was an established building type in late antiquity. Its name describes its defining feature, three semi-circular exedrae or apses. The basic idea was realized in many variations, large and small, roofed and unroofed, free-standing and embedded, centralized and longitudinal; and it was used for many purposes, public and private, secular and religious. Most often the three exedrae are aligned on cruciform axes, as at Sohag, but the axes are not always of equal length; one may be elongated, isolating the central conch in a focal position, and the apses are not always of the same size.
In the class of triconches the Sohag examples stand out for their monumentality and relatively good state of preservation. Its triconch sanctuary makes the Church of St. Shenoute, in particular, a major monument of early Christian architecture, and places Sohag in an international nexus of ambitious and innovative church designs as well as at the source of a significant regional tradition.
Some details of the design of both of the Sohag triconches are unknown or speculative due to damage and restorations. The triconch of the Church of St. Bishoi (the Red Monastery) is closer to its original state than the triconch of St. Shenoute (the White Monastery), which seems to have burned and partially collapsed at an early date. Both triconches were closed off from their respective naves to become independent buildings in the Middle Ages, and both are now still hidden behind tall north-south walls with small doors, as shown in the plan of the Church of St. Shenoute published by Monneret de Villard as the “etat actuel” of 1925 (fig. 1).
Plans and sections made by students at the Technische Hochschule Darmstadt in 1962 (figs. 2—4) show both triconches in essentially their present form, although some features have changed in the intervening forty-five years. Monneret’s axonometric rendering of the triconch of St. Bishoi (fig. 5) intimates the grandeur of conception that can still be sensed in both buildings, despite all of their trials and deformations. The impact is easier to grasp today at the Red Monastery thanks to the efforts of the conservators and other experts coordinated by Elizabeth Bolman.
Both triconches feature semi-circular exedrae opening off a square space. Passageways through the lateral apses lead east into rooms tucked into the spaces between the outer curves of the exedrae and the straight walls that frame the building externally, and west into transverse spaces between the triconch and the nave. The three apses are all of the same size and height, and all are similarly articulated with two rows of framed niches. In their present state, both triconches have domes over the central space and semi-domes over the apses.
The semi-domes in the triconch of St. Bishoi are original and convey the intended appearance. The domes of both triconches, however, are replacements for a superstructure of a different kind. Somers Clarke, who studied these buildings before Monneret de Villard and the restorations promoted by the Comite de Conservation des Monuments de l’Art Arabe, opined that because of the relative thinness of its walls the triconch of St. Shenoute must have been covered originally by a wooden roof, an opinion shared today by Peter Grossmann. The triconch of St. Bishoi, smaller and later than that of the White Monastery, may have been vaulted but probably was not domed.
Another significant difference between the triconches occurs in the relation of the exedrae to the central space. In the church of St. Shenoute’s monastery, the east apse opens directly off the central square but the north and south apses are separated from it by two broad arches, the second of which currently is higher and carries a small cupola (fig. 2). The effect is to elongate the square and to push the lateral apses outward into near alignment with the aisles of the basilica; in other words, to make the triconch occupy more of the width of the east end.
The triconch at the Red Monastery is more compact, with all three exedrae opening directly from the center; consequently, its width is not much greater than that of the nave. The compact design requires that the passageways to the peripheral spaces go through the walls of the north and south apses, rather than through piers under intermediate arches as in the Church of St. Shenoute.
The passageways in turn alter the pattern of the wall niches, as they replace two of them at ground level in each apse (fig. 3). Thus while the eastern apse has three niches, two square and one rounded, the lateral apses each have two rounded niches flanking a column. By contrast, the design of the triconch of the White Monastery church permits all three apses to be identical, with five alternately rectangular and semi-circular niches in each. The alternation is ingeniously managed so that a semi-circular niche falls in the center of the east apse and rectangular ones in the centers of the side apses.
The fourth side of both triconches was open to the west under an arch supported by tall columns. Clarke noted that the north-south brick wall that divides the triconch of St. Shenoute from its basilica must coincide with an original division of the church, because the pavement between it and the triconch—”formed of sundry slabs of red granite, bearing traces of hieroglyphs and patterns, terribly broken up”—was uniformly 38 cm higher than the pavement of the nave.
Not much more could be said about the western facade, however, because this zone of St. Shenoute’s church has been so thoroughly rebuilt. Clarke turned to the triconch of the Red Monastery church, where there was also a transverse wall (later removed) in which he could see four columns, a shorter pair aligned with the columns of the aisles of the basilica and a taller, inner pair that framed the opening into the triconch (fig. 5).
The central columns are thought to have supported what Peter Grossmann later termed a “forward triumphal arch.” Clarke also found original paving in this area, “consist[ing] of small squares of dark granite and basalt, inlaid upon bands of white marble.” These fragments were evidently still in situ in 1962 (fig. 3) but are not there today. As at the Church of St. Shenoute, there was a change of level between the space in front of the triconch and the nave, which was lower.
Literary and epigraphical evidence indicates that the Church of St. Shenoute, including the triconch, was built and dedicated just before 450 with the participation of “the most illustrious Count Caesarius, son of Candidianus,” who is referred to as “founder” in an inscription over the door into the south aisle. Recent attempts to date the Red Monastery triconch place it a century later, in the fourth or fifth decades of the sixth century.
There is some reason to think that the later triconch was originally free-standing; in any case it seems to have been built and decorated by more skilled workmen than the basilica. Clearly, the purpose of the Red Monastery triconch was to reproduce—duplicate or ‘copy’—the triconch of St. Shenoute’s church, for reasons that are still uncertain. The design was given by St. Shenoute’s triconch, and the genesis of the design must be sought in the context of the fifth century, historical and typological.
Monneret de Villard sought to answer the question of origins taxonomically, by sorting all of the triconch basilicas known in his day into three types: (A) aisled basilicas with a trefoil structure enclosed at the east end; (B) a simplified version in which the trefoil is appended to a single nave; (C), the type of the Church of St. Paulinus at Nola in Campania (fig. 6, G), which “according to texts” was an aisled basilica with a trichorum.
“A” is the type of the Sohag churches and the church in the precinct of the Hathor Temple at Dendera (now dated by Grossmann to the sixth century), as well as the monastery Church of St. Theodosius (Dayr Dosi) in Palestine and the ceremonial hall in the Umayyad palace at Mschatta in Syria. Monneret believed that the “A” group constituted an “Egypto-Palestinian” type that was the source of the “B” group, and “C” was a parallel invention. He argued that the type originated in Syria, or with a Syrian architect, because other aspects of the triconches also showed Syrian connections.
The triconch form itself and the elevation of the Sohag apses, with superimposed rows of niches, were a Roman inheritance; Monneret de Villard pointed in particular to the nymphaeum at Gerasa in modern Jordan. Triconches were widespread in Roman architecture and Monneret de Villard found that they were most common in baths and palaces. Although he did not make much of it, Monneret de Villard implied that it was through palace architecture that the trefoil form entered churches.
Twenty years later, Andre Grabar published the groundbreaking first volume of Martyrium, which offered a comprehensive explanation not only of the triconch but of all centralized forms in Christian architecture on the basis of what he called a “functional” approach: “the same religious functions call forth the same types of buildings and images.” To account for the many examples of cellae trichorae found in Christian cemeteries in Rome, North Africa, Gaul, Pannonia, and Palestine, Grabar looked for pagan buildings of like form and function. Although the archaeological record is weak, an inscription found in Tolentino commemorating a funerary pan- teum cum tricoro (also cited by Monneret de Villard) bolstered his case that Christian triconch chapels originated in the architectural typology of pagan tombs.
Grabar’s overarching theory was that Christians deliberately adopted the forms of pagan funerary architecture for martyria—buildings marking the tombs of martyrs as well as “theophanic” sites in Palestine—because of their functional associations. Thence these forms came into use for non- funerary purposes, as at Sohag. An additional factor specific to Sohag, in his view, was Shenoute’s intention to establish new pilgrimage centers in Upper Egypt, “to invite [Egyptian] Christians . . . to replace pilgrimage to the Holy Land with pious visits to the churches of their own country.
Shenoute . . . urged his countrymen to consider the grand monasteries that he founded at Sohag as Jerusalem, as sacred as the ancient city in Palestine.” This suggested that the triconch had become a signifier of ongoing theophany rather than of particular historic manifestations of the deity.
Grabar later observed that his theory of meaningful typologies resembles the linguistic concept of “semantic families.” Irving Lavin evoked a similar connection with his term “associative architecture” in an influential article of 1962, which sought to modify Grabar’s theory by allowing for the prominence of triconch rooms in palaces from late antiquity onward. “The triconch . . . appears as a kind of test case in the evolution of medieval ‘associative’ architecture . . . . Largely as a result of its associations, a Roman architectural tradition was transferred to Byzantium, there to play a role [in the typology of Byzantine palaces] that has no real parallel in the west.”
Extending his argument to ceremonial triclinia of all shapes, Lavin endorsed the long-standing belief that the aulic associations of centralized forms made them seem suitable for churches. “With the establishment of Christianity as the state religion, many of its official aspects were conceived in terms of the imperial cult . . . . The heavens are ‘imperial domiciles,’ and the eternal city a palace . . . . The church is also described in the same terms as the royal palace.” To signify the “house of the Lord,” church buildings were built to look like the grand rooms where earthly rulers received their guests.
Lavin did not apply the analogy of palace triclinia explicitly to triconch basilicas, but Peter Grossmann did so thirty years later. Grossmann proposed that there could have been a metonymic transfer from ceremonial triclinia of the type seen at Piazza Armerina to the site of the Christian eucharist (“since the holy communion in a church is in a way understandable as a kind of meal”), not directly, but via cemetery triconches where the eucharist was also celebrated.
Subsequently, he qualified this argument to account for the possibility that the idea for the triconch design came from Shenoute or his monks, not an architect. This might make a derivation from funerary triconches seem more likely; yet “it cannot be excluded that the triconch was viewed as an exceptionally rich spatial form, and for that reason was used in an ambitious church building . . . ; in that case only correspondingly shaped palace spaces would have come up as models, which takes us again back to dining rooms.”
Grossmann was responding to a challenge to the existence of such associative typologies raised by Tomas Lehmann, whose archaeological study of the buildings erected and described by St. Paulinus of Nola at Cimitile, northeast of Naples (fig. 6), led him to reconsider the origin of the triconch basilica. Following a number of German scholars more or less militantly skeptical of the idea that antique architectural forms were or could be ‘ Bedeutungstrager, Lehmann insisted that “like nearly every spatial form in Roman architecture, the centralized building was not tied to any purpose, nor was it in the early Christian period, when one finds centralized spaces in various forms as baptisteries, memorial buildings, and episcopal and community churches.”
Insisting further that the historical study of triconches should be based on modern taxonomic criteria rather than the imprecise vocabulary of ancient sources, Lehmann demonstrated that the building that St. Paulinus himself called absidem trichoram—the sanctuary space of his new basilica in the cemetery of St. Felix—is not a triconch, because the lateral “apses” are smaller and lower than the eastern one and do not, therefore, fit the model of “three, usually semi-circular conches or apses of equal size, which are added in the manner of a cloverleaf to a central crossing that is often a square” (fig. 6, G). Since Paulinus’s Basilica Nova was dedicated in 403, the effect of eliminating it from the category of triconch basilicas was to reopen the question of when and where the type originated; Egypt (Sohag), Crete, and North Africa are all contenders.
Applying an equally formalist approach, Iris Stollmayer argued that triconch basilicas were not a type (Baukonzept), so Lehmann’s exclusion of St. Paulinus’s trefoil-ended basilica is unfounded; this “supposedly homogenous class of buildings disintegrates into a variety of particular forms,” including the one at Cimitile. Stollmayer effectively takes us back to Monneret de Villard, positing three formally defined categories: freestanding triconches (Zentralbauten), single-naved triconch basilicas, and triconch basilicas with nave and aisles.
But unlike Monneret, however, she denied that these categories reveal any filiations or genealogy; on the contrary, they break down into discrete regional groups: Egyptian, Lycian, Armenian, Croatian. Only buildings within regions show direct interrelationships, with one church often serving as the model for the rest, as the Sohag churches were prototypical for Egypt. Stollmayer also denied any functional connotations of the triconch; according to her, because the form was employed in many contexts, including villas, baths, and cemeteries, it had no fixed associations.
The possibility of “semantic families” of buildings or of “associative architecture” is ruled out, at least for triconches. “The buildings cited are not related to one another by a standardized form, nor demonstrably by function or symbolism, nor can any common origin or genesis be discerned. Late antique triconch churches have in common a ‘theme,’ which . . . was varied at will . . . .”
The result of these latest interventions is to leave us with the proverbial elephant in the room: an imposing triconch basilica whose presence cannot be explained. It is not at all obvious that the monastery church of a desert father in Upper Egypt would be the first datable instance of this “theme,” as well as its most ambitious realization. If the idea was Shenoute’s, one feels compelled to ask where he got it, why it seemed appropriate, and how he communicated it to the workmen charged to construct it.
The last question points to the difficulty with Stollmayer’s position that the triconch basilica was not a Baukonzept. That the triconch, at least, was indeed an architectural concept is indicated by the existence of a word for it, trichorum. The triconch existed in discourse, like the basilica, and discourse was one means of its dissemination. Another was non-verbal representation in plans and models. Donors carry models of their buildings in early Christian mosaics (for example, in the apse of San Vitale at Ravenna), and the use of plans is attested for a building exactly contemporary with St. Paulinus’s constructions at Nola, the cathedral at Gaza erected in 402—407.
According to the fifth-century biographer of the cathedral’s founding bishop, as the people of Gaza were arguing about what form the new church should take, whether that of the destroyed temple of Zeus Marnas, which it replaced, or a totally different one, a letter arrived from the Empress Eudoxia.”Enclosed in the letter was the plan (oKapi^og) of the holy church in the form of a cross . . . and the letter contained instructions that the holy church be built according to this plan. . . . Furthermore, the letter announced the dispatch of costly columns and marbles.” Following these instructions,
The holy Bishop [Porphyry] … engaged the architect Rufinus from Antioch, a dependable and expert man, and it was he who completed the entire construction. He took some chalk and marked the outline (0eoig) of the holy church according to the form (o’/’npa tov OKap^ou) [circumflex should sit on top of ‘eta’, not beside it] that had been sent by the most pious Eudoxia.
A triconch plan etched into the pavement of the temple court at Dendera is in 1:1 scale to the triconch that actually was built there, so it is certain that the drawings were a factor in the spread of ecclesiastical triconches in Egypt.
The likelihood that other aspects of the Gaza scenario were reenacted at Sohag—the intervention of an important, if not imperial, secular patron and the participation of an architect from a major metropolis—supports the reconstruction of the process proposed by Monneret de Villard, although the architect did not necessarily come from Syria. The monumental niched triconch in the court of the Peirene Fountain in Corinth, recently redated from the second century to the second half of the fourth, suggests the possibility of late antique prototypes in the eastern Mediterranean, and, as has often been suggested, Alexandria may be a more likely source of the Sohag design.
Betsey Ann Robinson traced the Peirene triconch to the triclinia first studied by Lavin, and explained it as the deliberate appropriation of a building type from “domestic architecture of the most opulent sort,” a transferral to the public sphere of “an architecture of convivium” that would have been very familiar to members of the social class that governed Corinth in late antiquity.
That class was a frequent presence at Sohag. According to Heike Behlmer, “the number and high rank of the government officials Shenoute reports to have been intimate with is quite astonishing,” and included governors, counts, and military commanders. “From Shenoute’s own works one obtains the impression of a secular aristocracy disposed to travel . . . in style, accompanied by their families, clerks and servants . . . . They would come freely to the local holy man . . . and discuss theological questions among equals. . . .” Among them was the comes Caesarius, who came often, at least once with the hegemon Taurinus and their entourages, to discuss problems in the church at Alexandria.
If the “Magnificent Count Caesarius” commemorated as founder on the lintel in St. Shenoute’s basilica was not this Count Caesarius but another (which seems unlikely), he was in any case of the same social stratum. To such men, the grandiose form of the triconch would have seemed as appropriate as it did to the ruling class in Corinth—appropriate on one plane to their own status as patrons, and on another, allegorical plane to the Lord whose theophany was ongoing in the Christian liturgy that would be performed there.
It is possible to agree with Tomas Lehmann and Iris Stollmayer that the triconch had no fixed or exclusive associations without entirely rejecting the idea that triconches were a form of “associative architecture.” Peter Grossmann cautioned that similarities among buildings must be judged in three dimensions—in mass and height, kind and quality of decoration, “the quality of space”—rather than in the “calligraphy” of the ground plan.
It is from elevations that associations—we might think of them as social memories—arise. In elevation the Sohag triconches evoke certain public buildings—the fountains in Corinth and Gerasa. A different elevation might have recalled other kinds of triconches and therefore, a different set of associations.
In plan or concept, however, all triconches represent the number three. St. Paulinus made this obvious by decorating the main apse of his triconch with an image of the Trinity, accompanied by verses that are preserved in a letter to his friend Sulpicius Severus: “In full mystery sparkles the Trinity . . . . The holy unity of the Trinity meets in Christ, who likewise has His insignia in threefold . . . .” Paulinus described his triconch in the same terms, as a unity of three: “an undulating apse (absis sinuata) unfold[ing] itself with two recesses, one to the right and one to the left.” The symbolism of three was latent in every triconch, whether or not the patron recognized or exploited it. Whatever the associative meanings of the triconch’s material realization, it was the symbolic potential of the concept that evidently appealed to St. Paulinus, and arguably to St. Shenoute as well.
 Monneret de Villard 1925-1926: vol. 2: 48.
 Krautheimer 1986: 113-17.
 Grossmann 2002a: 533-34 dates the fire that partially destroyed the triconch to the Persian invasion in 619.
 Peter Grossmann kindly shared these drawings with Nicholas Warner, project architect at the Red Monastery, who in turn shared them with me.
 Clarke 1912: 153; Grossmann 2006: 45.
 Clarke 1912: 153.
 Clarke: 169.
 Lefebvre 1920a: cols. 471-475; Emmel 1998: 93-95.
 Severin (forthcoming).
 Grossmann 2002a: 538.
 For the church at Dendera see Grossmann 2002a: 443-446. For Dayr Dosi: Stollmayer 1999: 134 and 148 Nr. 30.
 Monneret de Villard 1925-1926, vol. 2: 129-30. For the nymphaeum: MacDonald 1986: 200 and Fig. 105 (p. 106).
 Monneret de Villard 1925-1926, vol. 1: 47-64.
 Grabar 1946, vol. 1: iii.
 Ibid.: 102-19.
 Ibid.: 328.
 Ibid.: iii.
 Lavin 1962: 12.
 Ibid.: 16.
 Grossmann 1992: 190.
 Grossmann, 2002b: 124.
 Lehmann 1996: 320-21.
 Ibid.: 323, 354.
 Stollmayer 1999: 140.
 Ibid.: 141.
 Grossmann 2002a: 445.
 Robinson 2001: 118-31. I am grateful to Sarah Lepinski for directing me to this study.
 Ibid.: 134-39.
 Behlmer 1998: 346-51.
 Grossmann 1992: 188.
 Paulinus of Nola, Epistula 32, 10.8—15: trans. Goldschmidt 1940: 38—39.
 Paulinus of Nola, Epistula 32, 13.14: trans. Goldschmidt 1940: 40-41.
 This study of the Sohag triconches was carried out under the auspices of the Red Monastery Project in collaboration with the Egyptian Supreme Council of Antiquities and the Coptic Church. I am grateful to Zahi Hawass, Abdallah Kamel, Magdi al-Ghandour, Abdallah Attar, and Muhammad Abd al-Rahim of the SCA, as well as to His Holiness Pope Shenouda III, Bishop Yohannes, Father Wissa, and Father Antonius for their generous hospitality and support. Work is being funded and administered by USAID and ARCE, under the Egyptian Antiquities Conservation Project (EAC) Agreement No. 263-A-00-04-00018-00. Thanks to Michael Jones, Red Monastery Project Manager for EAP/ARCE, and EAC Director for his assistance and on-site advice. I also thank Dr. Elizabeth Bolman for the priceless opportunity to be part of the Red Monastery team.